Abstract 5074: Profiling the splicing landscape in solid tumors in a large, real-world dataset

Abstract

Abstract Introduction Aberrant pre-mRNA splicing is common across cancers. Molecular regulation of splicing is an attractive drug target and clinical trials are currently evaluating splice-targeted therapies (STTs). Due in part to the complex and pleiotropic nature of splicing regulation, molecular biomarkers to identify patients who may benefit from STTs are lacking. While mutations in splicing factors (SFs) are promising biomarkers, these are rare, leaving many patients with unmet need. In this study, we identified and characterized splicing patterns (SPs) in a large, multi-cancer cohort. Methods De-identified patient records with splicing data from the Tempus xR whole transcriptome assay were extracted from the Tempus clinicogenomic database (N=2918). A total of 1052 alternative splicing events were detected in at least 90% of the samples with at least 10 supporting reads. Leiden clustering was applied to the percent spliced-in (PSI) values of these events to identify SPs. Fisher’s exact test was used to assess clinical and molecular enrichment in SPs, and Cox proportional-hazards models were used to assess associations with rwOS and derive hazard ratios (HRs). Results From a list of 404 SF genes 7.7% of patients had one or more pathogenic or likely pathogenic (P/LP) SF variants, and 2.2% had variants in the key biomarker SF genes SF3B1, SRSF2, or U2AF1. Clustering revealed 15 SPs containing 43 to 353 samples. One SP had a significant enrichment of SF variants (19.7% of SP, P < 0.0005). SPs were associated with significant differences in rwOS, and many were enriched for samples from a single cancer type, biopsy, or histology - principally by prostate cancer (97.5% of SP), melanoma (94.3% of SP), urothelial carcinoma (93.4% of SP), and renal clear cell carcinoma (93.0% of SP), as well as samples collected from liver biopsies (93.9% of SP). In contrast, colorectal and lung adenocarcinoma samples were present in a wide variety of SPs. One SP was enriched for squamous histologies (66.6% of SP, P < 0.0005) and associated with shorter rwOS (HR = 1.33, 95% CI [1.11, 1.59], P < 0.0005). This association remained after controlling for squamous histology (HR = 1.45, 95% CI [1.02, 2.05], P = 0.04). Samples in this SP were characterized by increased expression of MYC- and proliferation-related genes, and lacked enrichment for SF variants (P = 0.51, 6.6%). Conclusions Mutations in SFs were rare in our cohort, highlighting the need for additional biomarkers related to STTs. This work revealed the presence of multiple SPs - characterized by distinct clinical and molecular traits. Further work will be able to contextualize STT response data using SPs to facilitate STT biomarker discovery. Citation Format: Mario G. Rosasco, Andrew J. Sedgewick, Justin Guinney, Michelle M. Stein. Profiling the splicing landscape in solid tumors in a large, real-world dataset [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5074.